Can We Quickly Learn to “Translate” Bioactive Molecules with Transformer Models? DOI
Emma Tysinger, K. Brajesh, Anton V. Sinitskiy

et al.

Journal of Chemical Information and Modeling, Journal Year: 2023, Volume and Issue: 63(6), P. 1734 - 1744

Published: March 13, 2023

Meaningful exploration of the chemical space druglike molecules in drug design is a highly challenging task due to combinatorial explosion possible modifications molecules. In this work, we address problem with transformer models, type machine learning (ML) model originally developed for translation. By training models on pairs similar bioactive from public ChEMBL data set, enable them learn medicinal-chemistry-meaningful, context-dependent transformations molecules, including those absent set. retrospective analysis performance subsets ligands binding COX2, DRD2, or HERG protein targets, demonstrate that can generate structures identical most active ligands, despite having not seen any against corresponding target during training. Our work demonstrates human experts working hit expansion easily and quickly employ translate texts one natural language another, "translate" known given novel same target.

Language: Английский

Deep generative molecular design reshapes drug discovery DOI Creative Commons

Xiangxiang Zeng,

Fei Wang, Yuan Luo

et al.

Cell Reports Medicine, Journal Year: 2022, Volume and Issue: 3(12), P. 100794 - 100794

Published: Oct. 27, 2022

Recent advances and accomplishments of artificial intelligence (AI) deep generative models have established their usefulness in medicinal applications, especially drug discovery development. To correctly apply AI, the developer user face questions such as which protocols to consider, factors scrutinize, how can integrate relevant disciplines. This review summarizes classical newly developed AI approaches, providing an updated accessible guide broad computational development community. We introduce from different standpoints describe theoretical frameworks for representing chemical biological structures applications. discuss data technical challenges highlight future directions multimodal accelerating discovery.

Language: Английский

Citations

145
Generating 3D molecules conditional on receptor binding sites with deep generative models DOI Creative Commons
Matthew Ragoza,

Tomohide Masuda,

David Ryan Koes

et al.

Chemical Science, Journal Year: 2022, Volume and Issue: 13(9), P. 2701 - 2713

Published: Jan. 1, 2022

The goal of structure-based drug discovery is to find small molecules that bind a given target protein. Deep learning has been used generate drug-like with certain cheminformatic properties, but not yet applied generating 3D predicted proteins by sampling the conditional distribution protein-ligand binding interactions. In this work, we describe for first time deep system molecular structures conditioned on receptor site. We approach problem using variational autoencoder trained an atomic density grid representation cross-docked structures. apply atom fitting and bond inference procedures construct valid conformations from generated densities. evaluate properties demonstrate they change significantly when mutated receptors. also explore latent space learned our generative model interpolation techniques. This work opens door end-to-end prediction stable bioactive protein learning.

Language: Английский

Citations

123
Link-INVENT: generative linker design with reinforcement learning DOI Creative Commons
Jeff Guo, Franziska Knuth, Christian Margreitter

et al.

Digital Discovery, Journal Year: 2023, Volume and Issue: 2(2), P. 392 - 408

Published: Jan. 1, 2023

Link-INVENT enables design of PROTACs, fragment linking, and scaffold hopping while satisfying multiple optimization criteria.

Language: Английский

Citations

48
Equivariant 3D-conditional diffusion model for molecular linker design DOI Creative Commons
Ilia Igashov, H. Stärk,

Clément Vignac

et al.

Nature Machine Intelligence, Journal Year: 2024, Volume and Issue: 6(4), P. 417 - 427

Published: April 11, 2024

Abstract Fragment-based drug discovery has been an effective paradigm in early-stage development. An open challenge this area is designing linkers between disconnected molecular fragments of interest to obtain chemically relevant candidate molecules. In work, we propose DiffLinker, E(3)-equivariant three-dimensional conditional diffusion model for linker design. Given a set fragments, our places missing atoms and designs molecule incorporating all the initial fragments. Unlike previous approaches that are only able connect pairs method can link arbitrary number Additionally, automatically determines its attachment points input We demonstrate DiffLinker outperforms other methods on standard datasets, generating more diverse synthetically accessible experimentally test real-world applications, showing it successfully generate valid conditioned target protein pockets.

Language: Английский

Citations

42
Machine learning-aided generative molecular design DOI
Yuanqi Du, Arian R. Jamasb, Jeff Guo

et al.

Nature Machine Intelligence, Journal Year: 2024, Volume and Issue: 6(6), P. 589 - 604

Published: June 18, 2024

Language: Английский

Citations

39
Recent Advances in Automated Structure-Based De Novo Drug Design DOI Creative Commons
Yidan Tang, Rocco Moretti, Jens Meiler

et al.

Journal of Chemical Information and Modeling, Journal Year: 2024, Volume and Issue: 64(6), P. 1794 - 1805

Published: March 14, 2024

As the number of determined and predicted protein structures size druglike 'make-on-demand' libraries soar, time-consuming nature structure-based computer-aided drug design calls for innovative computational algorithms.

Language: Английский

Citations

27
3D molecular generative framework for interaction-guided drug design DOI Creative Commons
Wonho Zhung, H.G. Kim, Woo Youn Kim

et al.

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: March 27, 2024

Abstract Deep generative modeling has a strong potential to accelerate drug design. However, existing models often face challenges in generalization due limited data, leading less innovative designs with unfavorable interactions for unseen target proteins. To address these issues, we propose an interaction-aware 3D molecular framework that enables interaction-guided design inside binding pockets. By leveraging universal patterns of protein-ligand as prior knowledge, our model can achieve high generalizability experimental data. Its performance been comprehensively assessed by analyzing generated ligands targets terms pose stability, affinity, geometric patterns, diversity, and novelty. Moreover, the effective mutant-selective inhibitors demonstrates applicability approach structure-based

Language: Английский

Citations

23
In silico modeling of targeted protein degradation DOI Creative Commons
Wenxing Lv,

Xiaojuan Jia,

Bowen Tang

et al.

European Journal of Medicinal Chemistry, Journal Year: 2025, Volume and Issue: 289, P. 117432 - 117432

Published: Feb. 20, 2025

Language: Английский

Citations

2
RELATION: A Deep Generative Model for Structure-Based De Novo Drug Design DOI
Mingyang Wang, Chang‐Yu Hsieh, Jike Wang

et al.

Journal of Medicinal Chemistry, Journal Year: 2022, Volume and Issue: 65(13), P. 9478 - 9492

Published: June 17, 2022

Deep learning (DL)-based de novo molecular design has recently gained considerable traction. Many DL-based generative models have been successfully developed to novel molecules, but most of them are ligand-centric and the role 3D geometries target binding pockets in generation not well-exploited. Here, we proposed a new 3D-based model called RELATION. In RELATION model, BiTL algorithm was specifically designed extract transfer desired geometric features protein-ligand complexes latent space for generation. The pharmacophore conditioning docking-based Bayesian sampling were applied efficiently navigate vast chemical molecules with properties features. As proof concept, used inhibitors two targets, AKT1 CDK2. calculation results demonstrated that could generate favorable affinity

Language: Английский

Citations

63
Advances and Challenges in De Novo Drug Design Using Three-Dimensional Deep Generative Models DOI

Weixin Xie,

Fanhao Wang,

Yibo Li

et al.

Journal of Chemical Information and Modeling, Journal Year: 2022, Volume and Issue: 62(10), P. 2269 - 2279

Published: May 11, 2022

A persistent goal for de novo drug design is to generate novel chemical compounds with desirable properties in a labor-, time-, and cost-efficient manner. Deep generative models provide alternative routes this goal. Numerous model architectures optimization strategies have been explored recent years, most of which developed two-dimensional molecular structures. Some aiming at three-dimensional (3D) molecule generation also proposed, gaining attention their unique advantages potential directly drug-like molecules target-conditioning This review highlights current developments 3D combined deep learning discusses future directions design.

Language: Английский

Citations

51